After months of pilot studies to fine-tune the paradigms, two main pilot experiments were conducted, the first to tackle how individuals process information in groups, the second how they update information following group decisions (this experiment was also the project of a master student I supervised). These pilot studies allowed us to write two pre-registrations on the Open Science Framework with clear hypotheses. The main experiments were then conducted and disseminated in internal meetings and two European conferences. A scientific paper was written including both experiments, the paper is now a preprint on the open science framework has been submitted to a peer-reviewed journal and received a first round of reviews.
Moving on to a more detailed description of the experiments and their results, both studies employed a minimal group paradigm to establish various levels of group affiliations: first-level in-group, second-level in-group, and outgroup. A minimal group paradigm consists of creating groups by assigning people to groups based on a preference, a color, or other factors. In the conducted experiments, groups were formed based on participants' preference for one painting over another and their country of residence. Participants reported feeling more affiliated with the group of people who shared the country of residence and the painting preference (first-level in-group), as compared to the group of people sharing only the country of residence (second-level in-group). The least affiliation was reported for the group of people sharing neither the country of residence nor the painting preference.
The first study (N=380) involved a learning task, aiming to understand how participants process information collectively. Here, we applied a reinforcement learning framework for data analysis. In the second study (N=300), a perceptual task was employed to examine how participants update information based on a previously made collective decision.
Our findings reveal that being in a group, regardless of affiliation, alters the representation of the world by reducing immediate information processing and increasing confirmation bias. Group affiliation effects became prominent under uncertainty, with participants relying more on in-group decisions when exploring the task or when faced with more ambiguous information for updating. Notably, deciding with the out-group, as opposed to the in-group, approached the experience of deciding alone, potentially mitigating the impact of reduced outcome consideration within groups. These findings have been presented in conferences and various internal and external talks.
In the line of my work on collective decisions, I have published scientific papers on what motivates people to join group decisions, attributions of responsibility in groups to self and others, and the importance of groups and social alignment in behavioral change during the Covid-19 pandemic.
My project included a secondment part in a political science research center, where I brought my collective decision-making research from the cognitive science perspective to the field of political science. This involved discussions with interdisciplinary researchers and teaching a 12-week course on behavioral insights and political behavior. The course focused on how behavioral sciences can inform political decisions and policies, aligning directly with the scope of my fellowship project on group dynamics in collective decisions and political behaviors. I communicated my research and findings to students who can be directly involved in policy-making.
In parallel, I collaborated with colleagues on a paper on approaches to public policy and collective intelligence, incorporating my research on collective decisions and aspects of group affiliation, shared responsibility, and group performance. Toward the end of the fellowship, I therefore transitioned to more applied research and have now started a job outside academia, focusing on applied research to behavioral transitions, leveraging my background and knowledge in collective decisions.
